Convert a voicemail to text

This is the part worth reading before you upload anything, because it is where most voicemail converters waste your time. Weeve decodes M4A, MP3 and WAV. Those cover the two ways people normally get a voicemail out of a phone system: sharing it from an iPhone, which produces an M4A, and unified messaging that emails the voicemail to you as a WAV or MP3 attachment.

What does not work is AMR, and to a lesser extent 3GP. Some carrier voicemail exports and some Android voicemail apps save in those formats, and no amount of renaming the file changes the codec inside it. If that is what you have, convert a copy to WAV or MP3 with any audio converter first, then come back. Saying this here costs us nothing and saves you an upload that was always going to fail.

Weeve has a separate page for voice memos, and the two get confused constantly. A voice memo is a recording you made yourself, usually in the Voice Memos app, and it is yours from the start. A voicemail is a message someone else left for you, which arrives through a phone system and has to be got out of that system before anything can read it.

That difference is why this page leads with formats and the voice memo page does not. A memo you recorded is already an M4A sitting in an app you control. A voicemail might be an M4A, a WAV, an MP3, or an AMR file you cannot use, depending entirely on whose phone system it passed through.

It is not a carrier feature. Verizon, AT&T and most business phone systems offer their own voicemail-to-text that transcribes messages as they arrive, and if you want every future voicemail read to you automatically, that is the thing to turn on, not this.

This page is for the voicemail you already have and need in writing: the one you want to quote accurately, keep a record of, paste into a case note, or forward as text rather than audio. It handles one file at a time, and it handles it without sending the recording anywhere.

Voicemails carry things people would not put in writing. A client leaving instructions, a candidate declining an offer, a supplier admitting a delay, a phone number and a name in the same breath. The words themselves are often the record that matters.

Weeve decodes the file in your browser, so the audio is never sent to a server and no copy of it is stored anywhere but your own device. The only thing that downloads is the speech model.

The text comes back punctuated and capitalised, with word timestamps, which matters on a voicemail where someone reads out a number or an address quickly. Speaker separation is available but rarely needed here, since most voicemails are one voice.

Exports are plain text, Markdown and PDF, and they reflect the transcript as you edited it rather than the raw output.